Myrna van Lunteren <[email protected]> writes: > On Wed, Jan 21, 2009 at 6:17 AM, Rick Hillegas <[email protected]> > wrote: >> Knut Anders Hatlen wrote: >>> >>> Rick Hillegas <[email protected]> writes: >>> >>> >>>> >>>> Thanks, Knut! I notice that you have removed [DERBY-2991] from the >>>> 10.5.1 release-planning page. Does that mean you are concerned that it >>>> will not make it onto the 10.5 release train? >>>> >>> >>> No, I removed it from the list of new features because I considered it a >>> bug fix, not a feature. It's still mentioned if you follow the >>> HighValueFixCandidates link in the 10.5.1 Bug Fixing section. Feel free >>> to add it back if you think it belongs in the feature list or in some >>> other section of the planning page. >>> >>> >> >> Thanks for the explanation, Knut. I think that it's an important fix which a >> lot of users will be keenly interested in. So I hope that it is highlighted >> in the release notes. I'll leave it to our release manager, Myrna, to decide >> how she wants to flag important fixes like this. >> >> Thanks, >> -Rick >> > > I think I agree that this is a bug fix, not a feature... > It's a good point though - shouldn't this bug fix get marked as > needing a release note? Or am I misunderstanding and there is no > impact to existing applications?
If all goes well, the applications shouldn't see any functional differences, except that some situations that led to a deadlock before do not lead to a deadlock when it's fixed. However, since the mechanism that causes the deadlocks was introduced as an optimization for some types of queries, fixing it may have a negative effect on the performance of those queries. That may warrant a release note. I'm still in the process of investigating how big the performance impact is and how it can be minimized. The current numbers indicate that with the suggested fix queries get something between 10% higher performance and 10% lower performance, depending on the actual query. The feedback I've received so far is that this impact is too high, so it may be that we'll have to give up on this approach, and then I'm afraid I've run out of ideas. -- Knut Anders
